Real-Time Vehicle Recognition and Improved Traffic Congestion Resolution

An intelligent traffic management system (E-Traffic Warden) is proposed, using image processing techniques along with smart traffic control algorithm. Traffic recognition was achieved using cascade classifier for vehicle recognition utilizing Open CV and Visual Studio C/C++. The classifier was trained on 700 positive samples and 1140 negative samples. The results show that the accuracy of vehicle detection is approximately 93 percent. The count of vehicles at all approaches of intersection is used to estimate traffic. Traffic build up is then avoided or resolved by passing the extracted data to traffic control algorithm. The control algorithm shows approximately 86% improvement over Fixed-Delay controller in worst case scenarios.